avfilter/dctdnoiz: use 32-bit (float) operations instead of 64 (double) for DCTs
This makes the code about 1.5x faster without any noticeable difference in the output.
Showing
This diff is collapsed.
Please
register
or
sign in
to comment